Public CMS data · 2024 · 90867 + 90868 + 90869

TMS billing flags explained

Initial treatment and motor-threshold determination, subsequent treatments, and motor-threshold re-determination—with published Medicare services, patient bounds, payments and same-code peer evidence.

Sign in to download CSV

CMS cohort fetched 2026-11-07T23:23:37.341Z · All three code reads complete · 0 NPI matches in saved OIG LEIE

Estimated Medicare paid

$35.4M

No comparable prior value

Published services

239,355

No comparable prior value

Rendering providers

391

11 with peer flags

Summed provider patient bounds

9,019–15,259

Not unique patients across providers

Paid per service

$148

Mix-dependent; not used as a cross-code flag

Top 10 provider payment share

32.6%

Published fee-for-service market only

What each screening rule means

Services per patient above 99th percentile of peers

Flags a same-code/setting annual services-per-patient ratio strictly above the peer 99th percentile. Patients can overlap across settings. Repeated or maintenance treatments and panel mix can explain differences; no clinical ceiling is claimed.

Real CMS example: Vladimir Guevara Vazquez

Peer evidence

Screening score 100 · High; highest same-code/setting shared peer score, not a probability of fraud.

90867 · Office · Services per patient above 99th percentile of peers: 2.36 vs peer median 1.05 · 185 peers (Psychiatry).

90868 · Office · Services per patient above 99th percentile of peers: 77.11 vs peer median 26.25 · 327 peers (Psychiatry).

Medicare paid per service over 2x peer median

Flags Medicare payment per service strictly over twice the same-code/setting peer median. Geography, adjustments and coding can explain differences. This is never compared across the three different TMS codes.

No provider carries this flag in the loaded report.

Patient count above 99th percentile of peers

Flags the reported beneficiary count strictly above the same-code/setting peer 99th percentile. Large practices can legitimately treat more people; volume is not wrongdoing.

Real CMS example: Jorge Betancourt

Peer evidence

Screening score 75 · High; highest same-code/setting shared peer score, not a probability of fraud.

90867 · Office · Patient count above 99th percentile of peers: 175.00 vs peer median 15.00 · 185 peers (Psychiatry).

90868 · Office · Services per patient above 99th percentile of peers: 57.72 vs peer median 26.25 · 327 peers (Psychiatry).

90868 · Office · Patient count above 99th percentile of peers: 175.00 vs peer median 16.00 · 327 peers (Psychiatry).

What these codes cannot tell us

90868 is subsequent treatment, not treatment failure. 90869 is motor-threshold re-determination with treatment, not proof of a new course. Annual rows contain no diagnoses, treatment response, authorizations or claim dates. Organization NPIs and unusual specialties are context, not automatic points.

Sources and limitations

Medicare Part B fee-for-service only; Medicare Advantage, Medicaid and commercial claims are excluded. CMS suppresses provider-code rows with fewer than 11 beneficiaries. Patient counts overlap across codes and settings; bounds are not unique national patients. Annual aggregates cannot establish treatment courses, failure, medical necessity, duplicate visits or fraud. Coverage varies by Medicare contractor; no Spravato dosing thresholds are applied.

Peer groups use the same code and office/facility setting nationally, and the same CMS specialty with at least 20 providers; smaller specialties fall back to all providers in that code/setting. Score = positive ratio z-score × 12 + 18 per flag, capped at 100; a provider’s highest eligible code/setting score is shown. Year-to-year scores can move as peers change.